Soft Delay - A PT2399 delay (yes, another one) with CD4007 inside.

[anotherjim]

Just what everybody has been waiting for I'm sure...

What's special about it?
Within the CD4007:

The pre-amp is an N-MOS (in pin10, out pin12) common source. Inverting so output ends up true polarity after the wet/dry mix.

It has tails! A P-MOS (in pin1, out 2, control 3) used as a switch in the record path. A single MOSFET works ok for this because there is no drain-source body diode connection unlike discrete MOSFET's. The body diode for the P-channel is terminated to the supply pin 14. Why did I use the P-channel? I felt like it.

Soft clip input. The input to the delay goes thru a CMOS inverter (in pin6, out 8 & 13), which starts to progressively limit gain from very low levels giving some compression. If it clips at all, it should clip here rather than the PT2399's own input op-amp. This inverter is also acting as the Regen (feedback) mixer via the "Soft" path.

2 Regen (Feedback) paths. One way "Hard" goes straight back into the delay, the other "Soft" goes back via the input inverter. This is a log pot because the path impedances had to be different, and about the middle of pot travel, the 2 paths are equal out of phase at the PT2399 input and mostly cancel out feedback signal. So it's "centre off" for single repeat. Full "Hard" stops just short of runaway feedback. "Soft" will runaway at maximum.

EQ set by coupling caps takes a lot of bass out the echoes, so good repeat sustain in possible before runaway happens. It's a soft, almost ghostly effect I was aiming for.

PT2399 set-up is standard, so nothing to add to that. I've simplified the EQ to single pole low-pass feedback caps, as many do. At longer delays, you can hear conversion noise (you only hear it when you play). Increasing the low pass filtering might help, but makes it dull. I don't think you can win. If you want clean delay long enough to phrase loop, the PT2399 isn't the right puppy, unless you want dirty audio.
You may notice it all runs off 5V. I've not noticed a problem with that.

One flaw/feature is that when the MOS switch is "open", Soft Regen gets louder. I think this is due to the removal of "back load" from the preamp to the inverter input where the Soft regen is mixed. CMOS inverters don't seem to maintain a strong virtual ground at input, so each input resistor can somehow see the other input. If soft Regen is close to runaway, operating the switch to tails bypass can lead to runaway, and no way to stop it without turning the Regen down. Don't use strong Soft Regen unless you can deal with that. Alternately, fit a second true-bypass footswitch in the standard way or increase R7 to 100k if you don't ever want runaway feedback. Runaway is used in one of the demo tracks.

I have some audio demo's, but I made the mistake of using a sensitive condenser mic on the speaker - so you can hear the footswitch operating sometimes and a little pick "click". The tails switching is really silent - honest! No additional reverb was used.
You can hear the audio here...
https://soundcloud.com/ashdalestudio/sets/4007-soft-delay

.

[Marcvv]

Wow, impressive. I have to check this out a bit more later.

[stringsthings]

Sounds very good.  Are you going to make a PCB layout for this?

[anotherjim]

Sorry, I don't do pcb's. Mine is perfboard using a piece I already had which turned out to be 2 or 3 rows smaller than ideal.
Components all fitted but it need 6 or more wire jumpers to get some of the connections in!

I can supply the Eagle project by e-mail if anyone wants to work up a pcb, but I only use it to make a schematic. Eagle components were not chosen for a real board layout!

[vigilante397]

Very interested in the sound you've gotten here. I'm a huge fan of the PT2399 but the low headroom has always been a problem, and you seem to have solved it!I'd love to take a crack at a PCB, and my CNC machine has been gathering dust for almost a week I was going to ask for you to send me the files, but I don't even use Eagle so it would probably be just as easy for me to use the schematic you've posted above.

[alfafalfa]

Very interested in the sound you've gotten here. I'm a huge fan of the PT2399 but the low headroom has always been a problem, and you seem to have solved it!

Yes , this is by far the best delaysound I've heard from a PT2399 chip, much clearer than my Rebote 2.5.  Great job!


Alf

[anotherjim]

Nathan, go right ahead. Mine went in a 1590B with board mounted Alpha 9mm pots (first time I used those - excellent products). Would be easier in a BB, especially if you want true bypass.

Somehow, 4007 pinout never makes routing easy, no matter how it's used, something's always on the wrong side from where you want it.

Glad people like the sound. I've been living with it for a while and gotten used to it so what it has isn't so obvious to me now.

[Kipper4]

Sounds great Jim. Thanks for sharing.

[deadastronaut]

yeah, sounds nice, can never have too many delays....

[nickbungus]

Nice Jim!

[bluebunny]

yeah, sounds nice, can never have too many delays....

Tell that to the passengers on Southern Rail...   

[anotherjim]

yeah, sounds nice, can never have too many delays....

Tell that to the passengers on Southern Rail...   

Those aren't proper delays. Those are timetable improvements.

[Kipper4]

Might have to try this on the breadboard.

[anotherjim]

Might have to try this on the breadboard.

...well.....it....is....a.....delay.

[vigilante397]

So I had forgotten this thread existed as a lot can happen in 3 years I never got around to doing that PCB layout, and I still don't use Eagle, but my current software has an Eagle importer, so is it too late to take you up on that file? I could copy it from the picture but it's easier and more reliable to get it from a file.

[vigilante397]

Jim sent me the .sch file for this yesterday and I just finished the layout. I'm really glad I did this now instead of 2016 And Jim, hope you don't mind the dragon on the silkscreen. I figured give credit where credit is due

[anotherjim]

Oh wow -  that looks awesome. The Dragon is neat.
Good work.